For context, 5.3 Codex spark is 1000+ TPS so we are basically getting frontier intelligence at spark speed. The only issue w/ spark was the context window / compaction was atrocious compared to gpt 5.3/4/5 at 128k - hoping for atleast 256k context window
Folks aren't appreciating this enough - that is a 10-15x on gpt 5.5 fast mode and a 2x on gemini 3.5 flash (which is already atrociously fast (sans the BDE of gpt 5.5).
Folks aren't appreciating this enough - that is a 10-15x on gpt 5.5 fast mode and a 2x on gemini 3.5 flash (which is already atrociously fast (sans the BDE of gpt 5.5).
Even though Codex says its context window is only 272K tokens, I feel like it actually remembers everything I said in a session.
I'm using only one session for the past 2 weeks and it remembers stuff in the session way better than I do. 🤯
@thsottiaux Codex is GOATED. I would love to see a way to save queued items for later and not just dequeue, context is I realize I want to work on something first or I jumped the gun - I want to save it without having to delete it or save it locally somewhere. Esp annoying w/ screenshots
Every Mac can run up to 2 dedicated MacOS VMs.
Use Tart.
Install codex and computer use.
Turn on remote control.
Be done with the cloud vms or fake sandboxing policies.
As nature intended!
Excited for @OpenAIDevs to expand this to 1) support audio narrations 2) integrate with Chronicle for passively observed skill gen / up keep.
Show Codex a workflow once. Reuse it as a skill.
Record & Replay lets you show Codex a recurring task, like filing an expense report or submitting a time-off request.
Codex turns that demo into an inspectable, editable skill.
You control when recording starts and stops.